IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 47, NO.

2, MARCH/APRIL 2011

707

Transformer Failure Due to Circuit-Breaker-Induced Switching Transients

David D. Shipp, Fellow, IEEE, Thomas J. Dionise, Senior Member, IEEE, Visuth Lorch, and Bill G. MacFarlane, Member, IEEE

AbstractSwitching transients associated with circuit breakers have been observed for many years. Recently, this phenomenon has been attributed to a signicant number of transformer failures involving primary circuit-breaker switching. These transformer failures had common contributing factors such as the following: 1) primary vacuum or SF-6 breaker; 2) short cable or bus connection to transformer; and 3) application involving dry-type or cast-coil transformers and some liquid-lled ones. This paper will review these recent transformer failures due to primary circuitbreaker switching transients to show the severity of damage caused by the voltage surge and discuss the common contributing factors. Next, switching transient simulations in the electromagnetic transients program will give case studies which illustrate how breaker characteristics of current chopping and restrike combine with critical circuit characteristics to cause transformer failure. Design and installation considerations will be addressed, particularly the challenges of retrotting a snubber to an existing facility with limited space. Finally, several techniques and equipment that have proven to successfully mitigate the breaker switching transients will be presented, including surge arresters, surge capacitors, snubbers, and these in combination. Index TermsElectromagnetic Transients Program (EMTP) simulations, RC snubbers, SF-6 breakers, surge arresters, switching transients, vacuum breakers.

Fig. 1. Simplied electrical distribution system for data center.

which can fail distribution equipment such as transformers. Transformer failures due to circuit-breaker-induced switching transients are a major concern, which is receiving attention in a draft standard [1] and the focus of this paper. A. Forensic Evidence for a Unique Case Consider the case study of a new data center with a 26-kV double-ended loop-through feed to six dry-type transformers each rated at 3000 kVA AA/3390 kVA FA and 26/0.48 kV and deltawye solidly grounded, as shown in Fig. 1. The transformer primary winding is of 150-kV basic impulse insulation level (BIL). A vacuum breaker was used to switch the transformer. The 4/0 cable between the breaker and the transformer was 33 kV and 133% ethylene propylene rubber. Primary arresters were installed. The transformers were fully tested, including turns ratio, insulation resistance, etc. Functional tests were completed, including uninterruptible power supply (UPS) full load, UPS transient, data center room validation, etc. In the nal phase of commissioning, a pull-the-plug test was implemented with the following results. 1) De-Energization Failure #1. Four electricians simultaneously opened four 26-kV vacuum breakers to simulate a general utility outage. All systems successfully transferred to standby generation but a loud pop was heard in Substation Room B and the relay for the vacuum circuit breaker feeding transformer TB3 signaled a trip. 2) Energization Failure #2. Minutes later, two electricians simultaneously closed two 26-kV vacuum breakers to substation Room A. Transformer TA3 failed catastrophically.

I. I NTRODUCTION ODAY, medium-voltage metal-clad and metal-enclosed switchgears that use vacuum circuit breakers are applied over a broad range of circuits. These are one of many types of equipment in the total distribution system. Whenever a switching device is opened or closed, certain interactions of the power system elements with the switching device can cause high-frequency voltage transients in the system. The voltagetransient severity is exacerbated when the circuit breaker operates abnormally, i.e., current chopping upon opening and prestrike or reignition voltage escalation upon closing. Such complex phenomena in combination with unique circuit characteristics can produce voltage transients involving energies

Manuscript received June 4, 2010; accepted July 9, 2010. Date of publication December 23, 2010; date of current version March 18, 2011. Paper 2010PPIC-188, presented at the 2010 IEEE Pulp and Paper Industry Conference, San Antonio, TX, June 2123, and approved for publication in the IEEE T RANSACTIONS ON I NDUSTRY A PPLICATIONS by the Pulp and Paper Industry Committee of the IEEE Industry Applications Society. The authors are with Eaton Electrical Group, Warrendale, PA 15086 USA (e-mail: daviddshipp@eaton.com; thomasjdionise@eaton.com; visuthlorch@ eaton.com; billgmacfarlane@eaton.com). Color versions of one or more of the gures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identier 10.1109/TIA.2010.2101996

0093-9994/$26.00 2010 IEEE

708

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 47, NO. 2, MARCH/APRIL 2011

TABLE I H ISTORY OF T RANSFORMER FAILURES R ELATED TO P RIMARY VACUUM B REAKER S WITCHING

Fig. 2. Transformer failure #2 during energization.

Fig. 3. Transformer failure #1 during de-energization.

Failure #2 is shown in Fig. 2. Examination of the primary windings revealed that the coil-to-coil tap burnt off and the winding terminal showed an upward twist. The burn marks from the initial Flash indicated the transient concentrated on the rst turns of the windings. Typically, closing the vacuum breaker to energize the transformer is the worst condition. Failure #1 is shown in Fig. 3. Examination of the primary windings revealed Flash and burn marks on the B-phase winding at the bottom and middle. Those at the top indicate a coilto-coil failure, not a winding-to-winding failure, and indicate a transient voltage with high dv/dt. Those in the middle were a result of the cable (used to make the delta connection) swinging free. Supports were only lacking for this jumper (oversight during manufacturing) which could not withstand the forces of the transient. This transformer passed the BIL test at 150-kV BIL but ultimately failed at 162-kV BIL. All six transformers and cables were identical, but only two failed during the vacuum-circuit-breaker switching. The signicant difference was that the two failed units had 40 ft of feeder cable while the others had 80 or 100 ft of feeder cable. This short 40-ft cable, high-efciency transformer, and vacuum circuit breaker proved to be the right combination to produce a damaging voltage transient on both energization and de-energization. B. History of Failures and Forensic Review The previous example is not an isolated case. Instead, it is representative of a growing number of transformer failures

due to primary switching of vacuum breakers. Table I details a history of transformers related to primary switching of vacuum breakers occurring within the past three years. In Case 1, in a hydro dam, the transformer was value engineered with a 13.8-kV primary-winding BIL of 50 kV. The BIL should have been 95 kV for the 13.8-kV class. The 1955 switchgear was replaced with modern vacuum breakers with only 20 ft of cable to the transformer. The user chose to energize the transformer before conducting a switching transient analysis and failed the transformer primary winding. The post mortem analysis revealed that no surge protection was applied. In Case 2, in a hospital, the vacuum breaker was close-coupled through 27 ft of cable to a 2500-kVA dry-type transformer with a 95-kV-BIL primary winding. The vacuum breakers were supplied with no surge protection because the particular vacuum breaker installed had a very low value of current chop. During vacuum breaker switching of the transformer, the transformer failed. The transformer was rewound, and surge protection/snubbers were installed. In Case 3, in a railroad substation, vacuum breakers applied at 26.4 kV were used to switch a liquid-lled rectier transformer with 150-kV-BIL primary winding. The switching transient overvoltage (TOV) failed the middle of the primary winding. Forensic analysis determined a rectier with dc link capacitors, and the transformer inductance formed an internal resonance that was excited by the switching. Such an LC series resonance typically fails the middle of the transformer primary winding. In Case 4, in a data center, vacuum breakers applied at 26.4 kV were used to switch six dry-type transformers with 150-kV-BIL primary windings under light load. Two transformers failed, one on breaker closing and the other on opening. The failed transformers were connected by 40 ft of cable to the vacuum breaker, while the other transformers had either 80 or 100 ft of cable. Arresters were in place at the time of failure, but there were no snubbers. In Case 5, in an oil eld, a dry-type transformer for a variable speed drive (VSD) had multiple windings to achieve a 36-pulse effective harmonic-free VSD. A vacuum breaker at 33 kV was separated from the transformer by only 7 ft of cable.

SHIPP et al.: TRANSFORMER FAILURE DUE TO CIRCUIT-BREAKER-INDUCED SWITCHING TRANSIENTS

709

TABLE II C URRENT C HOP V ERSUS C ONTACT M ATERIAL

Arresters were applied on the primary winding. However, upon closing the breaker, the transformer failed. Finally, in Case 6, in an oil drilling ship, vacuum breakers designed to International Electrotechnical Commission (IEC) standards were applied at 11 kV and connected by 30 ft of cable to a dry-type cast-coil propulsion transformer rated at 7500 kVA. The transformer was also designed to IEC standards, and the primary winding had a BIL of 75 kV. The IEC transformer BIL is much lower than the American National Standards Institute (ANSI) BIL for the same voltage class winding. The transformer failed upon opening the breaker. C. Common Parameters The severity of the voltage surge, i.e., high magnitude and high frequency, and the damage caused by the voltage surge are determined by the circuit characteristics. The following are some rules-of-thumb to screen applications for potentially damaging switching transient voltages. 1) generally, short distance between circuit breaker and transformer (about 200 ft or less); 2) dry-type transformer (oil lled and cast coil not immune) and low BIL; 3) inductive load being switched (transformer, motor, etc.); 4) circuit-breaker switching characteristics: chop (vacuum or SF-6) or restrike (vacuum). D. Underlying Concepts 3) Current Chop. When a vacuum breaker opens, an arc burns in the metal vapor from the contacts, which requires a high temperature at the arc roots [2]. Heat is supplied by the current ow, and as the current approaches zero, the metal vapor production decreases. When the metal vapor can no longer support the arc, the arc suddenly ceases or chops out. This chop out of the arc called current chop stores energy in the system. If the breaker opens at a normal current zero at 180 , then there is no stored energy in the system. If the breaker opens, chopping current at 170 , then energy is stored in the system. Current chop in vacuum circuit breakers is a material problem. Older vacuum interrupters (VIs) used copperbismuth. Modern VIs use copperchromium. Most copperchromium VIs have a low current chop of

Fig. 4. Voltage escalation due to successive reignitions.

35 A, offering excellent interruption performance and a moderate weld strength. Table II shows the average and maximum levels of current chop for copperchromium, copperbismuth, and other contact materials. It should be noted that both vacuum and SF-6 interrupters current chop. Current chop is not unique to vacuum breakers. 4) Reignition. Current chop, even though very small, coupled with the system capacitance and transformer inductance can impose a high-frequency transient recovery voltage (TRV) on the contacts. If this high-frequency TRV exceeds the rated TRV of the breaker, reignition occurs. Repetitive reignitions can occur when the contacts part just before a current zero and the breaker interrupts at high-frequency zeros, as shown in Fig. 4. On each successive reignition, the voltage escalates. The voltage may build up and break down several times before interrupting. Although current-chop escalation with modern VIs is rare, a variation of this concept applies on closing called prestrike. 5) Switching inductive circuits. The transformer is a highly inductive load with an iron core. The effect of switching this inductive load and core must be considered. The current cannot change instantaneously in an inductor. Energy cannot be created or destroyed; only the form of energy is changed. The energy in the inductor is described by 1/2LI 2 = 1/2CV 2 or V = I L/C. (1)

From the energy equation, it can be seen that, for short cables, C is very small, which results in a very high surge impedance L/C . Energizing a cable produces a traveling wave which reects when it meets the discontinuity in surge impedance between the cable and the transformer. The surge impedance of a cable may be under 50 , while the surge impedance of the transformer is 3003000 . In theory, the reection can be as high as 2 per unit.

710

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 47, NO. 2, MARCH/APRIL 2011

Fig. 5. Important circuit elements for EMTP modeling.

Vacuum circuit breakers are prone to current chopping and voltage reignition while SF-6 circuit breakers are more prone to just current chopping. Air circuit breakers are not prone to either of any signicant magnitude. Manufacturers design vacuum-circuit-breaker contacts to minimize the severity and occurrence of abnormal switching leading to severe voltage surges (the lowest current-chop characteristics are 35 A). Regardless of the circuit-breaker manufacturer, voltage surges do occur. E. Characteristics of the Voltage Transient and Transformer Limits The voltage transient that develops following the vacuumcircuit-breaker switching is inuenced by three factors: stored energy, dc offset, and the oscillatory ring wave. The voltage component is due to the stored energy. The dc offset is determined by the X/R ratio of the cable and the transformer. The oscillatory ring wave is a result of the capacitance and inductance of the cable and the transformer. The magnitude of the voltage transient is compared with the transformer BIL. If the magnitude is excessive, then the transformer winding will likely fail line to ground. If the voltage transient has excessive rate of rise (dv/dt), then the transformer winding will likely fail turn to turn (natural frequency of ring wave). For the transformer to survive the transient, the insulation must be able to withstand both the magnitude and the dv/dt. Dry-type transformers are particularly susceptible to vacuum or SF-6 breaker switching transients. However, oil- or liquid-lled transformers are not immune. The oil has capacitance and acts like a surge capacitor to slow the rate of rise of the voltage transient. The trend in modern-day power systems is to install transformers with high-efciency design. As a result, these highefciency transformers have a very small resistance, which offers little or no damping to the voltage transient. Also, the repetitive effect, i.e., small indentations in the insulation, can occur with each successive peak of the voltage transient. II. P REDICTING P ERFORMANCE W ITH S IMULATIONS A. Modeling the Circuit When a statistical approach is taken for switching transients, complex modeling requires a frequency-dependent transformer model and an arc model of the circuit breaker. For purposes of screening applications for potentially damaging switching transients, a simpler approach is suggested with the important
Fig. 6. Simplied electrical system for ship propulsion drive.

circuit elements modeled in electromagnetic transients program (EMTP) consisting of the source, breaker, cable, and transformer, as shown in Fig. 5. The cable is represented by a Pi model consisting of the series impedance and half of the cable charging at each end. In some cases, multiple Pi models are used to represent the cable. The vacuum or SF-6 breaker is represented by a switch with different models for opening (current chop), restrike (excessive magnitude of TRV), reignition (excessive frequency of TRV), and closing (prestrike). The three-phase transformer model consists of the leakage impedance, magnetizing branch, and winding capacitances from high to ground and low to ground. For oil-lled transformers, the oil acts like a dielectric so the high-to-low capacitance is modeled. In cases requiring more detail, the transformer saturation and hysteresis effects are modeled. The choice of the integration time step will depend upon the anticipated frequency of the voltage transient. If it is too large, the time steps will miss the frequency effects. If it is too small, then this will lead to excessive simulation times. The Nyquist criteria call for a minimum sample rate of twice the anticipated frequency. In switching transients, the anticipated frequency is 325 kHz. When the circuit breaker opens, the transformer primary winding is ungrounded. Also, the ring wave is a function of the natural frequency of the circuit fnatural = 1/(2 LC ). (2)

The iron core of the transformer dominates the inductance of the circuit. The capacitance is very small for the dry-type transformer and short cable. Consequently, the circuits natural frequency is 325 kHz with relatively short cables.

SHIPP et al.: TRANSFORMER FAILURE DUE TO CIRCUIT-BREAKER-INDUCED SWITCHING TRANSIENTS

711

Fig. 8. TRV leading to reignition during energization of drive transformer with and without snubber. Fig. 7. Matching the simulation to eld measurements. TABLE III C URRENT C HOP AND R EIGNITION C ASES FOR D RIVE P ROPULSION T RANSFORMER S WITCHING

B. Mitigating the Switching Transient Various surge protection schemes exist to protect the transformer primary winding from vacuum-breaker-switchinginduced transients. A surge arrester provides basic overvoltage protection (magnitude only). The arrester limits the peak voltage of the transient voltage waveform. The surge arrester does not limit the rate of rise of the TOV. A surge capacitor in combination with the surge arrester slows down the rate of rise of the TOV in addition to limiting the peak voltage but does nothing for the reection or dc offset. The number of arrester operations is greatly reduced because of the slower rate of rise. There is a possibility of virtual current chopping. Finally, adding a resistor to the surge capacitor and surge arrester provides damping, reduces the dc offset of the TOV waveform, and minimizes the potential for virtual current chopping. The resistor and surge capacitor are considered an RC snubber. Selecting the values of resistance and capacitance are best determined by a switching transient analysis study, simulating the circuit effects with and without the snubber. C. Matching the Model to Measurements The results obtained from simulation of switching transients in EMTP are only as good as the choice of model and data used. When available, eld measurements taken during the switching transients enable verication of the EMTP model. The EMTP model can be adjusted as needed to match the actual eldmeasured conditions. To illustrate this approach, consider the ship propulsion electrical system in Fig. 6. The system consists of 3 2865-kW generators, a 4160-V three-phase bus, two 1865-kW drives/motors for forward

propulsion and identical drives for reverse propulsion, eight 1185-kVA dry-type transformers, and eight 630-A vacuum circuit breakers. The critical parameters are the vacuum circuit breaker, 50 ft of cable, and dry-type transformer of 30-kV BIL. Fig. 7 shows that the EMTP simulation results match the transients captured in the eld with a high-speed power-quality meter (closing). The simulation shows 4.96 kVpeak , which is less than 30-kV BIL; however, the oscillation frequency of 20.2 kHz exceeds an acceptable limit of dv/dt. Having veried the model, a series of current-chop cases and reignition cases were run. Fig. 8 shows the TRV leading to reignition and the TRV with a snubber installed. Reignition occurs because the TRV peak, time to crest, and rate of rise of recovery voltage exceed IEEE ANSI C37.06 limits.

712

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 47, NO. 2, MARCH/APRIL 2011

Fig. 9. Simplied electrical distribution system for Tier III data center.

The snubber reduces the TRV below the IEEE/ANSI limits for general-purpose vacuum breakers [3] and for generator breakers [4]. Table III summarizes the reignition cases and the current-chop cases. In all cases, the snubber is effective in reducing the transient voltage.

D. Borderline Case It is important to note that not all applications involving primary switching of transformers using vacuum breakers require snubbers. The large majority of applications do not require snubbers. Switching transient studies are conducted to determine when snubbers are needed. In this paper, the cases were selected to show different situations requiring snubbers. For the system shown in Fig. 9, the results were borderline; therefore, a snubber was still applied for reliability purposes. The Fig. 9 system is a Tier III data center with two 24.9-kV incoming lines, two 12.5-MVA 25/13.2-kV transformers, a 13.2-kV ring bus, two 2250-KW generators, and six 3750-kV cast-coil transformers. Data centers fall into the highest risk categories because of their high load density, close proximities of circuit components, highly inductive transformers (high-efciency designs), and frequent switching. The critical parameters for the Fig. 9 system are vacuum circuit breakers, 90-kV-BIL transformers, and cable lengths ranging from 109 to 249 ft. For the cable of 109 ft, the results of opening the vacuum breaker with current chopping of 8 A are shown in Fig. 10. The TOV is as high as 123 kVpeak on phase A which exceeds the transformer BIL of 95 kV. The TOV exhibits a signicant dc offset because there is very little resistance in the highly inductive circuit. The oscillation frequency of 969 Hz is slightly less than the acceptable limit. A snubber is required to reduce the peak below 95-kV BIL. The results of adding a snubber are shown in Fig. 10. Note the signicant reduction in the dc offset. The resistor in the snubber provides the reduction in dc offset as well as damping. The peak is reduced to 28.6 kV and an oscillation of 215 Hz,

Fig. 10. TOV initiated by current chop during de-energization of cast-coil transformer with and without snubber.

both within acceptable limits. Finally, eld measurements were taken after the snubber was installed to ensure that the snubbers performed as designed. The eld test setup for the snubber performance measurements is discussed in Section IV. The eld measurements showed that the snubber limited the TOV within acceptable limits. E. Case of Switching a Highly Inductive Circuit Now, consider the vacuum breaker switching of a highly inductive circuit, such as the starting current of a large grinder

SHIPP et al.: TRANSFORMER FAILURE DUE TO CIRCUIT-BREAKER-INDUCED SWITCHING TRANSIENTS

713

Fig. 12. TOV during de-energization of LMF transformer with and without snubber protection.

Fig. 11. Simplied electrical distribution system for LMF.

motor or an electric arc furnace. The vacuum-circuit-breaker switching of an electric arc furnace and ladle melt furnace transformers raises concern because of their high inductive currents. High-frequency transients and overvoltages result when the vacuum breaker exhibits virtual current chop and multiple reignitions. As an example, the arc furnace circuit of Fig. 11 consists of a 50-MVA power transformer, 2000-A SF-6 breaker, 56-MVA autoregulating transformer, 1200-A vacuum breaker, and 50-MVA furnace transformer. The switching of the SF-6 and vacuum breaker was studied. The vacuum breaker, because of the 28-ft bus to the furnace transformer, was the worst case. The results opening the vacuum breaker with and without snubbers are show in Fig. 12. The TOV of 386 kVpeak exceeds the transformer BIL of 200 kV, and the oscillation of 1217 Hz exceeds the acceptable limit. Application of the snubber results in a TOV of 56.4 kVpeak that is below the transformer BIL, and the oscillation of 200 Hz is below the acceptable limit. The results for cases involving current chop and reignition are given in Table IV. F. Concerns for the Pulp and Paper Industry The previous examples illustrate that circuit-breaker-induced switching transients can fail transformers for specic combinations of circuit parameters and breaker characteristics.

The examples show that the problem is not unique to one industry, application, vendors breaker, or transformer design. For the pulp and paper industry, there are many situations where circuit-breaker-induced switching transients are likely to damage transformers. The following examples are some of the more common scenarios encountered in the pulp and paper industry. 1) Vacuum breaker retrot for primary load break switch in a unit substation. In the pulp and paper industry, there are numerous unit substation installations with primary load break fused switch and no secondary main breaker. This arrangement results in arc Flash issues on the lowvoltage secondary. Limited space on the low-voltage side prevents installation of a secondary main breaker to mitigate the arc Flash issues. Retrotting a vacuum circuit breaker in the primary of the unit substation, in place of the primary load break switch, and sensing on the secondary is a solution that provides both primary and secondary fault protection [5]. Unit substations may have oil-lled or dry-type transformers. The secondaries may be solidly grounded or resistance grounded. With the vacuum breaker closely coupled to the transformer, surge arresters and snubbers are most likely needed. 2) Vacuum breaker and rectier (or isolation) transformer installation. Rectier transformers are installed to serve dc drives such as those needed for feed water pumps to the boilers. Also, isolation transformers are installed to serve a large VSD or groups of smaller drives. Primary voltages may be 13.8 or 2.4 kV, and secondary voltages

714

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 47, NO. 2, MARCH/APRIL 2011

may be 600 or 480 V. In both situations, vacuum breakers are installed in the primary and closely coupled to the transformer through a short run of bus or cable. Often, these transformers are inside and of dry-type design. 3) New unit substation with primary vacuum breaker. Recently, a paper mill installed a new metal-enclosed vacuum switchgear and a new 13.8/2.4-kV 7500-kVA transformer for a bag house for the generator boilers to meet Environmental Protection Agency requirements. The vacuum breaker was connected to the transformer through 5 ft of bus. While doing the coordination and arc Flash studies, the switching transient issue was identied. The equipment was installed and was awaiting startup and commissioning when the studies raised the concern. Before energizing the transformer, snubbers were quickly sized, obtained, designed, and installed. The screening criteria previously mentioned identify the aforementioned examples for potential damaging switching transient voltages due to vacuum breaker switching. The vacuum breaker, short distance to transformer, and dry-type transformer (or aged oil-lled transformer) are key variables to consider. With such short distance between breaker and transformer, most of these installations will require snubbers. One might conclude that standard snubbers could be applied. However, a switching transient study is still recommended to determine the unique characteristics of the circuit and custom design the snubber for the application. Given the limited space in each of these examples, it is unlikely that off-the-shelf standard snubbers would t. A substantial part of the design effort includes determining how to best t the snubbers into the new or existing unit substation or transformer enclosure. III. D ESIGNING THE S NUBBER The preceding analysis has shown that, in some cases, switching transients can produce overvoltages that can result in equipment insulation failure. If the results of the switching transient study indicate a risk of overvoltage greater than the BIL of the equipment and/or if the dv/dt limits are exceeded, a surge arrester and snubber should be applied. The switching transient study may also indicate that multiple locations require surge arresters and snubbers to protect the generator, transformer, or large motor. Additionally, the study species the necessary protective components and determines how close the protection must be placed to provide effective protection. A. Design Requirements At this point, custom engineering design determines how to best provide the protection needed for the equipment. The following questions must be answered to ensure that the snubber design meets all criteria and specications. 1) Is the switching transient protection cost effective? 2) What is the value of the equipment being protected? 3) What is the cost of lost production if the equipment fails from switching transients? 4) Can the protection be installed within existing equipment enclosures?

Fig. 13. Typical snubber and arrester arrangement for transformer protection.

Fig. 14.

Example of standard snubber package for transformer.

5) Will equipment modication void equipment warranties? 6) Are there standard equipment packages that can accomplish the switching transient protection? 7) Is the application an indoor or outdoor application? 8) Is there available space to mount the protective equipment in an external enclosure? 9) How can it be veried if the switching transient protection components are working effectively? 10) What alarms are necessary if the protective equipment components fail? B. Custom Engineering Design Fig. 13 shows the typical snubber arrangement for transformer protection. A noninductive ceramic resistor and a surge capacitor are the basic components of a snubber design. Resistance values typically range from 25 to 50 . Standard capacitor ratings that range from 0.15 to 0.35 F are the basis of the design. Fig. 14 shows a standard 15-kV surge protection package. The arresters are mounted on the top of the enclosure. A threephase surge capacitor is mounted on the bottom. Insulators and bus are located in the center. The cables can enter from top or bottom. A ground bus is located on the center right. If space heaters are required for outdoor locations, they are located on the lower left. Fig. 15 shows one phase of a custom snubber circuit. The custom design was required because there was not enough room in the transformer for the snubber components. The enclosure

SHIPP et al.: TRANSFORMER FAILURE DUE TO CIRCUIT-BREAKER-INDUCED SWITCHING TRANSIENTS

715

TABLE IV C URRENT C HOP AND R EIGNITION C ASES FOR LMF T RANSFORMER S WITCHING

Fig. 15. 15-kV snubber mounted above the transformer.

had to be mounted above the transformer. The cable connections from the transformer were eld installed and land on the copper bus. A 15-kV nonshielded jumper cable was used to make the connection. Each phase passed through an insulation bushing to the transformer below. Bus work was required to provide a solid support for the fragile resistors. Normally, only one resistor would be provided, but for this application, to achieve the delivery schedule, parallel resistors were designed to obtain the correct ohmic value (the correct single resistor value had long delivery). Fig. 16(a) and (b) shows a snubber assembly mounted in medium voltage switchgear. The photo on the left shows the single-phase surge capacitors mounted vertically. The black cylinders are ceramic resistors. A variety of options are available to detect if the snubbers are functional. They range from nothing (oversized but treated like a lightning arrester) to very sophisticated loss of circuit detection. Glow tube indicators are shown at the top of Fig. 16(a), and a close-up is shown in Fig. 17(a). These glow tubes are visible through a window in the switchgear door and provide a visual indication of snubber

Fig. 16. 15-kV snubber mounted in switchgear with glow tubes and current sensors.

continuity. The purpose of the blue current sensors at the bottom of Fig. 16(a) is to monitor the continuity of the resistor and fuse (optional) and alarm on loss of continuity. A close-up of the current sensor is shown in Fig. 17(b). Some industries mandate fused protection. If there should be a broken resistor or a blown fuse, an alarm signal can be sent to the plant distributed control system or supervisory control and data acquisition system to alert the operating personnel that these snubber components have failed. Fig. 16(b) shows a continuation of the same snubber assembly. Three fuses are attached to the tops of the resistor. The fuses will isolate any fault that may occur in the snubber assembly and prevent loss of the breaker circuit. C. Special Design Considerations The nature of high-frequency switching transients requires special design considerations. The snubber designer should

716

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 47, NO. 2, MARCH/APRIL 2011

Fig. 17. 15-kV snubber visual indication (glow tubes) and continuity verication (current sensors).

consider the location of the switching transient source when developing the custom design layout of the protective equipment. Abrupt changes in the electrical path should be avoided. A low inductive reactance ground path should be designed, using noninductive ceramic resistors and at tin braided copper ground conductors. The minimum clearances of live parts must meet or exceed the phase-to-phase and phase-to-ground clearances of NEC Table 490.24. The enclosure should be designed to meet the requirements of IEEE Standard C37.20.2 1999. When the enclosure is mounted greater than 10 ft from the equipment to be protected, NEC tap rules may apply to the cable size required and additional circuit protective devices may be required. D. Custom Designs for the Pulp and Paper Industry As mentioned previously, given the limited space in each of the examples related to the pulp and paper industry, it is unlikely that off-the-shelf standard snubbers would t. Instead, a substantial part of the design effort includes determining how to best t the snubbers into the new or existing unit substation or transformer enclosure. Following are three examples of the custom design effort needed for snubber installation. 1) 13.8-kV Solidly Grounded System in Paper Mill. A vacuum breaker was retrotted into the enclosure for the primary load break switch of the unit substation with a dry-type transformer. The space for the snubber was extremely limited, as shown in Fig. 18 (left). Because the system was solidly grounded, the voltage on the surge capacitor was limited to 8 kV line to ground; therefore, it was possible to use a three-phase surge capacitor. The tight clearance required the use of glastic to insulate the components at line potential from ground. 2) 13.8-kV Resistance Grounded System in Paper Mill. Another example of retrotting a vacuum breaker for a primary load break switch with a 30-year-old oil-lled transformer. Because this snubber was needed immediately, the only available resistors had to be paralleled to obtain the desired resistance, as shown in Fig. 18 (right). Again, glastic was used for insulation and to support the resistors. 3) 13.8-kV Low Resistance Grounded System. Snubbers were provided for a new metal-enclosed vacuum switchgear and a new 13.8/2.4-kV 7500-kVA transformer

Fig. 18. Snubber with (left) three-phase and (right) single-phase surge capacitors for 13.8-kV paper mill application.

Fig. 19.

Snubber for metal-enclosed 13.8-kV vacuum breaker.

for a bag house. Single-phase surge capacitors were used. Single resistors of the right ohmic value were available. Adequate clearance did not require the use of glastic as shown in Fig. 19.

IV. M EASUREMENTS TO V ERIFY S NUBBER P ERFORMANCE Following the installation of the snubbers, power quality measurements may be taken to ensure the proper operation of the snubbers. A high-speed scope or power quality disturbance analyzer should be used to measure the TOV waveforms at the transformer primary produced during switching of the primary vacuum circuit breaker. The measurements are used to verify that the waveforms do not exhibit excessive high-frequency transients (magnitude, rate of rise, and frequency). The test measurement setup generally consists of voltage dividers and a transient recording device. The voltage dividers should be made of capacitive and resistive components with a bandwidth of 10 MHz. The scope or power quality meter should be capable of transient voltage wave shape sampling,

SHIPP et al.: TRANSFORMER FAILURE DUE TO CIRCUIT-BREAKER-INDUCED SWITCHING TRANSIENTS

717

snubber to an existing facility with limited space. Finally, the performance of the snubbers is veried with eld measurements at the medium-voltage primary winding of the transformer. R EFERENCES
[1] ANSI/IEEE, A Guide to Describe the Occurrence and Mitigation of Switching Transients Induced by Transformer and Switching Device Interaction. C57.142-Draft. [2] D. Shipp and R. Hoerauf, Characteristics and applications of various arc interrupting methods, IEEE Trans. Ind. Appl., vol. 27, no. 5, pp. 849861, Sep./Oct. 1991. [3] Standard for AC High-Voltage Generator Circuit Breakers on a Symmetrical Current Basis, C37.013-1997, 1997. [4] Application Guide for Transient Recovery Voltage for AC High-Voltage Circuit Breakers, C37.011-2005, 2005. [5] D. Durocher, Considerations in unit substation design to optimize reliability and electrical workplace safety, presented at the IEEE IAS Electrical Safety Workshop, Memphis, TN, 2010, Paper ESW2010-3.

Fig. 20. Snubber performance measurement setup for 15-kV transformer using high bandwidth voltage dividers.

Fig. 21. Voltage divider connections at 18-kV surge arrester on transformer primary bushing.

8000-Vpeak full scale, and 200-ns sample resolution (5-MHz sampling). An outage and lockout/tagout are needed to install the voltage dividers, make connections to each of the three phases at transformer primary bushing, and make secondary connections to the transient recorder. Fig. 20 shows the test measurement setup for a typical cast-coil transformer. Voltage divider connections to the transformer primary were made, as shown in Fig. 21. Additionally, for tests requiring load at a transformer, a portable load bank may be connected to the transformer secondary. A resistive load bank congurable to different load levels (300 or 100 kW) prevents destructive testing. V. C ONCLUSION This paper has reviewed recent transformer failures due to primary circuit-breaker switching transients to show the severity of damage caused by the voltage surge and discuss common contributing factors. Next, switching transient simulations in EMTP were presented to illustrate how breaker characteristics of current chopping and restrike combine with critical circuit characteristics to cause transformer failure in unique situations. In these limited instances, mitigation of the transients is accomplished with snubbers custom designed to match the specic circuit characteristics. Design and installation considerations were addressed, particularly the challenges of retrotting a

David D. Shipp (S72M72SM92F02) received the B.S.E.E. degree from Oregon State University, Corvallis, in 1972. He is a Principal Engineer with the Electrical Services and Systems Division, Eaton Corporation, Warrendale, PA. He is a distinguished scholar in power system analysis and has worked in a wide variety of industries. He has spent many years performing the engineering work associated with his present-day responsibilities, which include a wide range of services covering consulting, design, power quality, arc ash, and power systems analysis topics. Over the last few years, he has pioneered the design and application of arc-ash solutions, modifying power systems to greatly reduce incident energy exposure. He has written over 80 technical papers on power systems analysis topics. More than 12 technical papers have been published in IEEE Industry Applications Society (IAS) national publications and two in EC&M. He spent ten years as a professional instructor, teaching full time. He occasionally serves as a legal expert witness. Mr. Shipp is currently the Chair for the IEEE Industrial and Commercial Power Systems-sponsored Working Group on generator grounding. He has received an IEEE IAS Prize Paper Award for one of his papers and conference prize paper awards for six others. He is very active in IEEE at the national level and helps write the IEEE Color Book series standards.

Thomas J. Dionise (S79M82SM87) received the B.S.E.E. degree from The Pennsylvania State University, University Park, in 1978, and the M.S.E.E. degree with the Power Option from Carnegie Mellon University, Pittsburgh, PA, in 1984. He is currently a Senior Power Systems Engineer with the Power System Engineering Department, Eaton Corporation, Warrendale, PA. He has over 27 years of power system experience involving analytical studies and power quality investigations of industrial and commercial power systems. In the metal industry, he has specialized in power quality investigations, harmonic analysis and harmonic lter design for electric arc furnaces, rectiers, and variable-frequency drive applications. Mr. Dionise is the Chair of the Metal Industry Committee and a member of the Generator Grounding Working Group. He has served in local IEEE positions and had an active role in the committee that planned the Industry Applications Society 2002 Annual Meeting in Pittsburgh, PA. He is a Licensed Professional Engineer in Pennsylvania.

718

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 47, NO. 2, MARCH/APRIL 2011

Visuth Lorch received the B.S.E.E. degree from Chulalongkorn University, Bangkok, Thailand, in 1973, and the M.S. degree in electric power engineering from Oregon State University, Corvallis, in 1976, where he was also a Ph.D. candidate and was inducted as a member of the Phi Kappa Phi Honor Society. During his Ph.D. studies, he developed the Short Circuit, Load Flow, and Two-Machine Transient Stability programs. The Load Flow program has been used in the undergraduate power system analysis class. He also prepared a Ph.D. thesis on the Load Flow program using the third-order Taylors series iterative method. In 1981, he joined Westinghouse Electric Corporation, Pittsburgh, PA. He was responsible for conduction power system studies, including short circuit, protective device coordination, load ow, motor starting, harmonic analysis, switching transient, and transient stability studies. He also developed the Protective Device Evaluation program on the main frame Control Data Corporation supercomputer. In 1984, he joined the Bangkok Oil Renery, Thailand, where his primary responsibility was to design the plant electrical distribution system as well as the protection scheme for the steam turbine cogeneration facility. He was also responsible for designing the plant automation, including the digital control system for the plant control room. In 1986, he rejoined Westinghouse Electric Corporation. He performed power system studies and developed the Short Circuit and Protective Device Evaluation programs for the personal computer. In 1998, he joined the Electrical Services and Systems Division, Eaton Corporation, Warrendale, PA, where he is currently a Senior Power Systems Engineer in the Power Systems Engineering Department. He performs a variety of power system studies, including switching transient studies using the electromagnetic transients program for vacuum breaker/snubber circuit applications. He continues to develop Excel spreadsheets for quick calculation for short circuit, harmonic analysis, soft starting of motors, capacitor switching transients, dc fault calculation, etc.

Bill G. MacFarlane (S70M72) received the B.S.E.E. degree from The Pennsylvania State University, University Park, in 1972. He began his career in 1972 with Dravo Corporation, Pittsburgh, PA, as a Power System Design Engineer supporting this engineering/construction company with expertise in the design and construction of material handling systems, chemical plants, pulp and paper mills, steel facilities, ore benefaction, mine design, and computer automation systems. He provided full electrical design services in heavy industrial applications, primarily in the steelmaking and coal-preparation-related sectors. His responsibilities included client and vendor liaison and supervision of engineers, designers, and drafters. From 1978 to 2003, he was a principal Process Controls Systems Engineer with Bayer Corporation, Pittsburgh. He provided design and specication of instrumentation and control systems for chemical processes. He congured programmable logic controller and distributed control systems. He was a Corporate Technical Consultant for power distribution systems. He had been the principal Electrical Engineer on many major projects with Bayer Corporation. He joined the Electrical Services and Systems Division, Eaton Corporation, Warrendale, PA, in 2004. At Eaton, he has been involved in a 230-kV substation design, ground grid design for substations, as well as in short circuit, protective device coordination, load ow, and arc-ash hazard analysis studies. He has also done power factor correction and harmonics studies to resolve power quality issues.